SolutionSoft
Sponsor Message

     

Hidden Value details commands and procedures in MPE that can improve your productivity with HP 3000 systems. Get a free NewsWire HP 3000 “3000 for 2000” cap — submit your MPE tip directly to us here at the NewsWire. Send your tips to editor@3000newswire.com, or fax them to 512.331.3807.

Edited by John Burke

We are running Apache version 1.2.5 on our HP 3000. When accessing a Web page on this server, the first page that loads takes about one minute to load, and every page loaded after that seems to load pretty quick (less than five seconds). All files are in bytestream format with little to no graphics on them. The same files placed on a non-HP 3000 Web server load almost instantly. How can I improve the response time?

Rick Gilligan and Mark Bixby reply:
Check to see if HostnameLookups is on. If so, try turning it off. Suspect it’s trying a reverse DNS (IP -> Name) to record in the log files.

Mark Bixby adds:
Another workaround is to make sure that your 3000’s /etc/resolv.conf, aka RESLVCNF.NET.SYS, is configured properly to point at a real DNS server.

I just received the following error messages from a query serial read:

>f invoice-number = “ “ and date-received ib 990604,990805
USING SERIAL READ
Bad SYSTEM LOGICAL Sector number 0x7A5FD0 on ldev 4 (io_status=FFDD0071)
Filename of bad file:/JOBSCOPE/DATABASE/INVENT48
<— snip —>
Bad SYSTEM LOGICAL Sector number 0x7A6090 on ldev 4 (io_status=FFDD0071)
Filename of bad file:/JOBSCOPE/DATABASE/INVENT48
FREADDIR FAILURE 48 96

Is this a bad drive? What is the “best” way to recover from this (other than a complete restore from last night’s backup)?


Stan Sieler replies:
It’s a hardware problem with the disk drive. Normally, HP would say “replace the drive, then reload.”
If you push, you should be able to get them to help you by connecting a replacement drive of the same size, (offline) cloning the old drive to the new drive, and then detaching the bad drive and booting up. (There are several ways to clone a drive, one using a free HP tool, one using Lund’s De-Frag/X.)

Note that this will probably leave you with bad data in some sectors of some files that were on ldev 4. We know that the sectors 0x7A5FD0, etc., are bad, so after you come back up with the new disk (after the clone), you’ll want to determine which files are on the bad sectors, and restore those files from the backup. In this case, you know the bad file is INVENT48.DATABASE.JOBSCOPE.

What you don’t know is what other sectors might have had bad data. That’s the argument in favor of a reload, because it doesn’t risk getting bad data from disk.

On the other hand, with the clone mechanism, you get up and running faster. You can then (somehow) build a list of all files that have any parts on ldev 4 and restore just that list. (Various tools/products can assist in this.)

In general, I go with the reload approach, but I’ve had a few sites where they couldn’t afford the downtime and wanted to take the risk with the cloning approach.

So, the answer to “best” is: It depends.

I’ve been using FCOPY to view spoolfiles. But some of them are too wide to fit on the screen. What other tools can I use?

Larry Barnes, Tony Knowles and Peter Chong reply:
After switching your terminal or terminal emulator into 132 character mode, you could use the print command. Print Onnnn.out.hpspool where “nnnn” is the spoolfile number. You can also add start=;stop=;and page= modifiers.

The print command can be combined in a command file with:

ECHO ![CHR(27)]&w6f132X to switch into 132-column mode, and
ECHO ![CHR(27)]&w6f80X to switch back to 80-column mode.

Is there any way to list the files on an MPE/iX store tape without restoring?

Several people suggested VSTORE. This will work, but also verifies the tape in the process and can take quite a bit of time. If the person asking the question just wants a simple, quick list of the tape contents, then Lee Gunter provided the best answer:
If the tape was recorded with Native Mode :STORE, :restore *t;;listdir

Use HELP RESTORE to determine the complete functionality of the LISTDIR option.

[Editor’s note: this was also the answer to another thread asking how to determine when a STORE tape was created. The header of the report gives not only the creation date but also the STORE command itself.]

I use a network printer and when I change something in my network configuration file (NPCONFIG) I must always restart my system (^A shutdown). If I don’t restart the system, it doesn’t understand my new configuration. Why?

Harlan Lassiter and Doug Werth reply:
You really just need to STOPSPOOL/STARTSPOOL the printers affected by the NPCONFIG change to make it work. By rebooting the system you are implicitly doing a STOPSPOOL and STARTSPOOL.

I have just installed one of HP’s latest and greatest 18.2Gb drives. I have defined it as a user volume. How do I migrate data from the system volume set over to the user volume? We can’t figure out how to copy data, groups and accounts from the system volume set to the new user volume set.

Denys Beauchemin replies:
The simplest way would be to store the groups and accounts to be moved to the new user volume and then purge them from the system. Next, recreate the accounts and the groups on the new user volume set with:

:newacct xyz;yaddayadda
:newacct xyz;onvs=”user volume””

Next, create the various groups for the accounts with:

:newgroup abc.xyz;homevs=”user volume”
:newgroup abc.xyz;onvs=”user volume”

Finally, restore the files. They will go to the proper groups and accounts on the user volume set.

Please note that you have to issue the newacct command twice. Once to describe it fully and the second time to set the onvs parameter. You need to do the same thing to the newgroup command. There are UDCs available on HP’s Jazz Web server (jazz.external.hp.com) to help you manage these volumesets.

Gerald Dillard adds:
A utility that may help with the migration of accounts and groups to the user volume is BULDACCT in PUB.SYS.

I have had a user stuck logged on since last Thursday. I have tried aborting the session but the session continues to live. What can I do?

Andrew Schriber summarized (with some comments and re-ordering thrown in by yours truly):
• Check for unspooled printer in use by session, and perform abortio.

• Check for file access conflict, and clear blocking process.

• Use abortio and termdsm (or OpenView DTC Mgr) to clear up user port if a serial (DTC) device

• Use command NSCONTROL KILLSESS if a VT session (actually, be sure to use this BEFORE abortjob for best results)

• Cycle the network, using NSCONTROL stop and start (disrupts all VT sessions and network traffic)

• Reboot the machine (the only solution guaranteed to work, but also the most disruptive)

Lars Appel adds:
Getting a stack trace of the hung processes can also be helpful in getting ideas about the reason why it is “stuck” and what potential measures might help it resume/abort.

Does anyone know the command to put a DDS drive back online after it has finished writing without ejecting the tape and putting it back in?

Ted Ashton, Cory Black reply:
:DEVCTRL.MPEXL.TELESUP n,,,Online
where n is the LDEV of your tape drive.

Barry Lake adds:
If you surf to Allegro’s Web site (www.allegro.com/software/) , there is a FREE program called ONLINE which will do the same. The nice thing about ONLINE is that it waits for AVR before returning to the user.
[Editor’s note: I’ve used ONLINE (written by Stan Sieler) for years. Works great, lasts a long time.]

After creating several links with the NEWLINK command, is there a way to see all links in effect on the system and for each link, the complete file name referred to by the link? Is there a LISTLINK or SHOWLINK command?

Jeff Vance replies:
LISTFILE linkname,7 shows the target of the link.
LISTFILE /,7;seleq=[object=symlink] shows all symbolic links on your system.

Mark Bixby added:
The POSIX solution would be:
find / -type l | xargs ls -l

Could you give me a cookbook recipe for enabling telnet on MPE/iX 5.5?

Christian Lheureux replies:
Enabling incoming telnet is feasible on MPE/iX 5.5, with this recommendation: make sure you use at least the PowerPatch 1 version of the release. Incoming telnet is less than perfect in 5.5.

• If not already done, configure the SERVICES.NET.SYS file. If the file does not exist, a sample can be copied from SERVSAMP.NET.SYS. Then link this file from posix space under the name /etc/services.

• If not already done, configure the PROTOCOL.NET.SYS file. If the file does not exist, a sample can be copied from PROTSAMP.NET.SYS. Then link this file from posix space under the name /etc/protocols.

• If you wish to enable security features (allow this service, disallow that one, e.g.), you will also need file INETDSEC.NET.SYS and its link /usr/adm/inetd.sec.

• You will also need file INETDCNF.NET.SYS (can be copied from sample INCNFSMP.NET.SYS) and its link from posix space /etc/inetd.conf.

All these files are more or less self-documenting. They’re basically ASCII files that you can modify with the editor of your choice.

Then you start job JINETD.NET.SYS. Telnet is one of the internal facilities of the INETD daemon launched by this job. This should do it.

Is there a way to disable/enable various system logging events without doing a restart?

Mike Hornsby replies:
From Stan Sieler and compatriots at Allegro’s terrific web site (www.allegro.com/software/) , I think you want SYSLOG, a utility that displays and changes what items are being logged by system logging (to the LOG#####.PUB.SYS files).

I did the PowerPatch 6 update last night. Now, I don’t see any of the UDCs from my third-party packages. What has happened?

Al Nizzardini replies:
Initially you are told to rename the file “command.pub.sys.” At the end of the process you need to purge command.pub.sys then reverse the process you did above. If you forgot the second step, do it now. If you forgot the first, you need to RESTORE your command.pub.sys from a backup tape. 


Copyright The 3000 NewsWire. All rights reserved.